Your AI infrastructure, in one place.

Call OpenAI, Anthropic, and Gemini through an OpenAI-compatible chat interface. Select a model per request and stream responses with existing clients.

Chat Completions, Responses, and OpenAI embeddings endpoints are available. Provider-specific parameters and behavior are documented in the API guide.

Use managed access or bring your own provider keys. Configure credentials at project or user level and track usage in Ingram Cloud.

Managed access includes model charges and a platform fee. With your own keys, the provider bills the tokens directly and Ingram Cloud bills the platform fee. Storage is charged separately.

Inspect model calls, tool calls, token counts, and costs for each run. Attribute model spend to users and projects.

Set budgets to alert or block new runs when limits are reached. Usage reports also expose prompt-cache reads and writes.

Connect MCP servers, enable built-in tools, and configure approval rules. Ingram Cloud runs server-side model-and-tool loops.

Your MCP server executes your custom tools. Ingram Cloud handles invocation, credential storage, OAuth refresh, and pausing for approval.

Persist conversations and user-scoped memory. Upload documents into managed vector stores for retrieval, or call the embeddings endpoint for your own search pipeline.

Working memory and semantic recall are configurable. Vector stores handle document chunking and search; the raw embeddings endpoint returns vectors without storing them.

Integrate through the API, expose a hosted page, connect messaging channels, or schedule recurring runs. Manage configuration with Pulumi.

Supported channels include Slack, Telegram, WhatsApp, Discord, and email. Channel setup and credentials are configured per project.